AWS re:Invent 2025 - Building Agentic AI: Amazon Nova Act and Strands Agents in Practice (DEV327)

The Future of Agentic AI: Practical Applications and Innovations

The Evolution of Agentic AI

• Agentic AI has emerged as an exciting and rapidly evolving field, with a clear progression from low-agency, rule-based systems to highly autonomous agents.
• The future of Agentic AI is envisioned as an "agent economy" - a global neural network of interconnected, autonomously operating intelligent agents.
• This shift will require not just technological advancements, but also changes in governance, trust, and organizational readiness.

Foundational Infrastructure for Agentic AI

• Communication protocols like MCP and A2A are addressing the need for interoperability between AI agents and their tools.
• AWS is actively participating in the development of these standards to leverage its experience in distributed systems.

AWS Agentic AI Portfolio

• AWS offers a comprehensive Agentic AI portfolio with services organized into three layers: infrastructure, AI/agent development, and applications.
• The focus of this presentation is on the AI and agent development software and services, including SDKs for agents like Amazon NovAct and Strands Agents.

Building a Weather Information Agent with Amazon NovAct

• Demonstration of rapidly building an AI agent application to retrieve local weather information using Amazon NovAct and the MCP protocol.
• The agent autonomously scrapes data from the Hong Kong Observatory website and generates a screenshot of the weather forecast.
• Technical details on implementing the agent using NovAct and MCP are provided, including code examples and blog post references.

Developing Custom Agentic AI Applications with Strands Agents

• Challenges in building fully functional AI agents, including the need for input connectors, memory, reasoning frameworks, persona definition, and observability/guardrails.
• Introduction to Strands Agents, an open-source SDK from AWS that simplifies agent development by handling complex orchestration and leveraging state-of-the-art language models.

Demonstration: Creating a Math Animation Agent with Strands Agents

• Showcase of building an agent that can generate a polished mathematical animation from a natural language prompt using Strands Agents and the Manim library.
• Detailed analysis of the core implementation code, demonstrating the simplicity of building complex Agentic AI applications with the Strands Agents framework.

Multi-Agent Collaboration Patterns

• Overview of four key multi-agent collaboration patterns: Agent as Tool, Swarm, Graph, and Workflow.
• Explanation of how these patterns enable predictability, control, and efficient development of Agentic AI solutions.
• Practical examples of applying these patterns in real-world scenarios, such as customer service chatbots, incident response, data validation, and automated data pipelines.

Agentic AI in Finance: A Quantitative Trading Example

• Demonstration of building an Agentic AI system for quantitative trading, featuring an orchestrator agent that coordinates specialized agents for market data retrieval, strategy generation, backtesting, and result summarization.
• Technical details on implementing the agent-based architecture, including the use of AWS Agent Core Gateway for secure tool integration and AWS Agent Core Lambda for agent deployment.
• Example of how the multi-agent system can rapidly iterate on trading strategies and provide comprehensive analysis to quantitative researchers.

Conclusion and Key Takeaways

• AWS's comprehensive Agentic AI portfolio, including services and tools like Amazon NovAct, Strands Agents, and Agent Core, provides a robust framework for building advanced AI agent applications.
• The presented demonstrations and use cases showcase the power of Agentic AI in diverse domains, from weather information to financial analysis and educational content creation.
• Developers can leverage the multi-agent collaboration patterns and open-source Strands Agents SDK to build custom, efficient, and scalable Agentic AI solutions for their businesses.